Groundwater contamination vulnerability assessment using DRASTIC method, GSA, and uncertainty analysis

Groundwater is one of the most important water resources in arid and semi-arid regions, whose quality and quantity are variable. In this study, the vulnerability of Ardabil aquifer was investigated using a classic and modified DRASTIC model. The weights of the parameters used in the DRASTIC model are universal; however, according to the characteristics of each region, these weights can be variable and associated with uncertainty. Sensitivity analysis is an important tool in vulnerability analysis to identify influential parameters. Global sensitivity indices are the mathematical models that can be computed by Monte Carlo methods and used for estimating the influence of individual variables or groups of variables on the model output. In this study, the global sensitivity analysis (GSA) was used to determine the weights of the layers of DRASTIC model based on the regional data in ArcGIS software. Using new weights extracted with GSA, the modified DRASTIC map was obtained for 2011 and 2016. The classic DRASTIC model classified the study area into medium to very high vulnerability, while the modified model presented very low to moderate vulnerability. Based on the new weights of DRASTIC model, the vulnerability increased due to the rise in precipitation in 2016 compared with 2011. To compare the results, the uncertainty map was analyzed using the Monte Carlo method of both classic and modified models. The results showed that the modified DRASTIC model was in a better condition, and this map was classified into 3 classes including the ranges of 0–1, 1–2, and 2–3. In both models, the groundwater level and the topography layers had the highest and lowest weights, respectively. Generally, in the case of the uncertainty map, it is better to determine the new weight for each region separately.